A biomechanical comparison of two screw fixation methods in a Letenneur type I Hoffa fracture

Sixteen medium, left side, 4th generation synthetic composite femurs (model 3403, Sawbones, Vashon, WA) were utilized in the study. Each specimen was cut into a 18 cm distal segment and divided into two groups of 8 each. Then, the anatomic axis of the femurs were well-aligned in a 6 cm cylinder tube. Afterwards, the specimen was potted in a cylinder tube using anchoring cement (PMMA).

Each specimen was predrilled prior to the osteotomy to facilitate the subsequent anatomic reduction. Furthermore, a specific prefabricated drilling jig was utilized to ensure the consistency of guide pin placement. Initially, two 3.2 mm threaded pin were advanced under fluoroscopic guidance. After that, the screw length was determined with a direct measuring gauge. Finally, a 4.9 mm cannulated drill and a 6.5 mm tap were used to prepare the two screw holes.

In order to mimic the Letenneur type I Hoffa fracture, an oscillating saw was utilized to perform the lateral condylar osteotomy, which was inferiorly extend from the extraarticular condyle-shaft junction to the articular surface. The fracture line was parallel to the axis of posterior femoral cortex. Then, the fracture was reduced anatomically and secured with one of the two screw trajectories described below. All of the surgical procedures were performed by one surgeon under fluoroscopic guidance to ensure accurate implant length and position.

A materials-testing machine (AG-X; Shimadzu Corp., Tokyo, Japan) was utilized for the biomechanical tests.. The specimen potted in a cylinder tube was secured distally into a fabricated adjustable metal clamp (Fig. 3). Afterwards, a flat stainless steel plate connected to a 5000 N load cell was centered on the Hoffa fracture to apply an axial compression force. Initially, a 100 N preload was applied to the specimens at a speed of 2 mm/min. Then, the load was increased to 300 N at a speed of 10 mm/min. The slope in the linearly elastic region of the load-displacement curve was recorded and calculated as the axial stiffness. Second, a 10,000 cycle repeated cyclic loading test was applied to the specimen with a force ranging between 200 and 600 N (valley/peak) at a frequency of 1 Hz, and the interfragmental displacement was recorded, respectively, after 10, 100, 1000, and 10,000 cycles. Finally, destructive axial compression was loaded at a speed of 10 mm/min on each specimen until catastrophic failure occurred, which might result from screw dislodging, fragment cracking or cortex disruption. The loading protocol mainly followed a previous study [11]. The maximum load detected by the testing machine during the load-to-failure test was defined as failure load. In addition, all failure modes were recorded for the purpose of the analysis. There were two sensors firmly fixed by cement onto the fragment and intact distal femur respectively. All the interfragmental displacement during the static and cyclic loading was measured with a magnetic tracking system (Polhemus, Colchester, VT, USA), by which the real-time data was detected at a resolution of 0.03 mm.

Following the pilot study, in which three specimens in each group was conducted, an priori power analysis was performed using G*Power [24]. The effect sizes were calculated from the preliminary results and were 2.17 for the failure load and 1.18 for the stiffness. To provide a study power of 0.9 and α value of 0.05, the projected sample size needed with this effect size is approximately 8 for each group. The final data from the two groups was conducted using SPSS software (SPSS Version 17; SPSS Inc., Chicago, IL, USA). To determine the differences in axial stiffness and destructive failure load between the two types of screw trajectories, considering the small sample size, the Mann-Whitney U test was performed for the distribution-free equivalent statistic on the data analysis. As regards the cyclic displacement, considering that repeated measurements of cyclic displacement were made four times on each specimen, a 4-step General Linear Model Repeated Measures Test was performed. A confidence interval of 95% was used for all the parameters to determine significance.

The interfragmental fracture displacement after each specific cyclic loading was collected, where it was found that the value increased steadily during the cyclic test. In the group A, the interfragmentary fracture displacement after 10, 100, 1000, and 10,000 cycles was 0.69 ± 0.19 mm, 0.88 ± 0.23 mm, 1.12 ± 0.25 mm, 1.36 ± 0.40 mm, respectively. In the group B, the interfragmentary fracture displacement after 10, 100, 1000, and 10,000 cycles was 0.90 ± 0.48 mm, 1.04 ± 0.55 mm, 1.14 ± 0.62 mm, 1.29 ± 0.61 mm, respectively. There were no statistically significant differences between the two fixation constructs after each cycle.

All of the specimens completed the entire cyclic loading test. Afterwards, the specimens were loaded until catastrophic failure. Based on the load-displacement curve, all the specimens exhibited changes in the curve slope prior to catastrophic failure, which indicated loss of fixation or plastic deformation of the construct. Followed with the plastic deformation, the maximum load detected by the testing machine was defined as the ultimate failure load, which was 1214 ± 127 N in the group A and 1109 ± 156 N in the group B. There were no significantly differences between the groups (p = 0.172). The most common pattern of failure was screw cut through the cancellous bone progressively, after that, the cortex bone was ruptured while maximum failure load was detected. (Fig. 4).